问HN：人工智能研究是否比更大的语言模型更需要“世界模型”？
看来，生成假设不再是主要问题。更棘手的问题可能是持续收集真实世界的数据，构建准确的世界模型，并利用这些模型来指导实验。这是否是自动化研究的发展方向？
查看原文
It seems that generating hypotheses is no longer the main problem.
The harder problem may be continuously collecting real-world data, building accurate world models, and using them to guide experiments.
Is this where automated research is heading?